Bindu Bhargava Reddy Chintam bhargavchintam

Hi, I'm Bindu Bhargava Reddy Chintam

ML systems engineer · GPU performance · LLM inference · AI agents whose work can be checked
Research Scientist at the Research Foundation for SUNY · San Francisco Bay Area

I make large models train and serve faster, and I build AI agents whose work can be checked.

🔬 Now: Research Scientist at the Research Foundation for SUNY. I profile and tune LLM fine-tuning on NVIDIA B200 (Nsight, Triton and CUDA kernels, FSDP, QLoRA) and build deep-learning models for genomics.
🏗️ Before: ML engineer at A5G Networks (5G traffic forecasting, Databricks and Delta Lake pipelines) and at Cloudphysician (real-time ICU alerting on Kafka, PySpark and AWS); data engineer at InxiteOut.
🧪 On weekends: hackathon builds on agent security, computer use, video agents, rare-disease diagnosis and cryo-EM.
📄 7 peer-reviewed papers and 3 patents, listed on Google Scholar.

Featured projects

aureproof An open-source integrity layer for AI-agent evaluations. It checks that the grader works before a task counts, signs an evidence bundle anyone can re-score offline, and gates regressions with a pre-registered pass / fail / inconclusive rule. _{Python · Inspect AI · in-toto · Sigstore · Apache-2.0 · pre-alpha, contributors welcome}	tripwire A security checkpoint in front of every AI-agent tool call. It holds risky actions before they run, trips on decoy secrets, traces a poisoned input across the fleet, and proposes guardrails proven by replay before a human approves them. _{FastAPI · React · ClickHouse · MCP · Semgrep · built with Sripadha at the SF Tech Week Cyberdefense Hackathon}
computer-use-capabilities An LLM works out a task in a hostile legacy bank UI once. The run compiles into a typed, versioned capability that replays with no model in the loop, never commits twice, and hands the live session to a person when it must. _{Python · Playwright · Anthropic SDK · Pydantic · 139 tests · 15 replay scenarios, each run twice with identical results}	llm-inference The LLM inference stack written out in NumPy: KV cache, paged attention, continuous batching, speculative decoding and quantization. Every piece checks itself against a reference, and CI runs all 29 checks. _{Python · NumPy · MIT · contributors welcome}
gpu-kernels CUDA and Triton kernels for deep-learning primitives (tiled GEMM, warp-shuffle reductions, fused softmax, layernorm, flash attention), each checked against a NumPy reference, with notes on coalescing, tiling and the roofline. _{CUDA · Triton · NumPy}	VizAgent A video agent that sweeps street-camera and dashcam footage for pedestrian–vehicle conflicts, double-checks every clip, ranks the hotspots and drafts the work order a city traffic engineer could act on. _{NVIDIA Cosmos · YOLO11 · VAST · FastAPI · built with Sripadha V and Jacob Shrader at the VAST Builders Challenge}

Hackathon and team builds

	Zuzu: call a number, answer plain questions in your own language, and six agents on BAND fill in a USCIS form, with every value taken from a deterministic tool. Built with Gar Walsh. _{🏆 Best Use of BAND and Best Use of ElevenLabs, Prompt Driven Hackathon 2026}
	CryoClear: a real-time cryo-EM junk-triage copilot. It picks particles, flags junk as micrographs stream in, and learns from a scientist's corrections. I built the picking engine, the junk classifier and the active-learning UI. _{QBI Hackathon 2026, UCSF · open source (MIT)}
	Diageno: a rare-disease diagnostic recommendation prototype. From phenotype and genetic data it ranks a differential diagnosis, shows the evidence and uncertainty, and suggests the most useful next question. _{Harvard Rare Disease Hackathon 2026 · team of four}

Open source

NVIDIA-NeMo/Curator#2285: showed that the fuzzy-dedup tuning guidance for num_bands and minhashes_per_band was inverted under MinHash-LSH banding. The maintainers fixed the docs in #2287.
spack/spack-packages#7083: makes charliecloud 0.44 and later require squashfuse 0.4.0 or newer (fixes #5400). In review.
promptdriven/pdd#2341: context limits are handled as a hard cutoff and provider prompt caching is never used; with an RFC on serving assumptions in #2342.
COMPAS-Lab/cloud-manager: 15 merged pull requests to a research lab's cloud console, a fork of Linode Cloud Manager (TypeScript, React).

Notes and practice

high-level-design: system design notes and case studies, including ML and AI-infrastructure designs with capacity estimates
low-level-design: 48 runnable object-oriented designs, from a parking lot to an inference server, a KV-cache pool and a GPU memory allocator
leetcode: 935 solutions in Python, C++, Java, C, SQL and bash
Study path: my B.Tech coursework repos, arranged as a curriculum

Panda3D games and QGIS maps

_{Dune Buggy Sandbox}

_{Orbital Defense}

_{Beam Bender}

_{Still Life Painter}

qgis-geospatial-projects: six QGIS and Python analyses (urban heat island, flood risk, crop health, healthcare access, forest change and solar siting). Each has QGIS layer styles, a Python pipeline and a Streamlit app, and the maps are live on the project site.